convert_crhome_bookmarks.py

from bs4 import BeautifulSoup
from fire import Fire


def bookmarks_to_links(html):
    soup = BeautifulSoup(html, 'lxml')
    links = [x.attrs['href'] for x in soup.findAll('a')]
    return links

def links_to_bookmarks(links):
    start = """<H1>Bookmarks</H1>
    <DL><p>
    <DT><H3 ADD_DATE="0" LAST_MODIFIED="0" PERSONAL_TOOLBAR_FOLDER="true">Bookmarks Bar</H3>
    <DL><p>
    </DL><p>
    <DT><H3 ADD_DATE="0" LAST_MODIFIED="0">Links</H3>
    <DL><p>
    """
    end = """
    </DL><p></DL><p>"""

    return start + "\n".join('<DT><A HREF="' + x + '">' + x + '</A>'  for x in links)

def main(input_filename, output_filename):
    with open(input_filename) as f:
        with open(output_filename, 'w') as w:
            if input_filename[-4:] == "html":
                html = '\n'.join(f.readlines())
                links = bookmarks_to_links(html)
                w.write('\n'.join(links))
            else:
                links = f.readlines()
                html = links_to_bookmarks(links)
                w.write(html)


if __name__ == "__main__":
    Fire(main)